Weird Verbal Scores on actual GMAT

In the past few weeks, some of us have given the GMAT and got surprising verbal scores. Most of these are either 27/28 .
Im unsure how reflective these scores are of our capbilities in the Vebal section

Similiar scores have also appeared across other forums.. Just wanted everyone who has given their GMAT on gmatclub in recent days to post their Verbal test experience , and what they think has gone wrong, so that there can be a better understanding on whats gone wrong...

Present here I took the test last Monday and had 27 in Verbal : i honestly thought there was a bug in the software!!!

Do yo feel you goofed up the Verbal Section, , were the CR's tough, or the Sc's difficult. Did you get any long science pasaages..

Although the bug issue sounds weird, way back in 2001, GMAT had a similiar issue ...

I took the exam before 2 weeks and Math was 48 and verbal was 27. Everything was looking fine till i saw the result.But 27 was out of Imagination.

Do you have any details on the 2001 issue?

Honestly i think i was doing decent during the Verbal section. The CR was as tough as in my prep when i was scoring around 38; i didn't notice it was easier maybe just a couple of them.
As for the SC, it was the same story not easy but not extremely difficult. I even remember getting un entire underlined SC around question 5 or 6 after my first RC passage.
One of the 4 RC passages was pretty long and the questions not trivial!!
I have done some 7-8 CAT tests and the OG to assert, i think, fairly the difficulty level.

I took the exam 2 days ago and the scored 28 on verbal. Even in my first GMATPrep I didnt score that low. I dont know what happened, but it just happened..

The link below speaks about the error. It happened in 2000 and not 2001.
http://www.fairtest.org/gmat-error-hurt ... -10-months

That's an interesting article. I don't think the author is very knowledgable about the world of GMAT, b-school apps, or b-school in general. The example the author cites is of college seniors. We all know that college seniors are not the most likely candidates for b-school as most b-schools want their students to have some work experience. An author knowledgeable in this would not have used college seniors as an example. The errors were by ETS, the company that used to proctor the GMAT exam. Now it's Pearson VUE. While Pearson is not immune to test errors, it is less likely due to the number of years since the first errors and the fact that GMAC should learn from its mistakes.

Additionally, the error was in 3% of test takers' experiences. It seems that what is being said here is that the possibility of this affects a large number of people. In the end, the scariest part is that all of you could potentially be right and that your scores did get messed up. I think there is a small change that this is actually what happened, but it's possible.

Since the low scores were in verbal, would those that experienced the lower than expected verbal state whether you are a native english speaker?

As usual, your analysis in this discussion is fine. I'm not a native english speaker though. That's an angle of analysis i did not take in account.

I am a native speaker. I did lower than I expected. However the test is designed to make you think you are doing well when you are really not. I hope they periodically review exams manually as a test to ensure accuracy of results.

terp26 - I think you're right. They should do some QA for all of us and periodically check some tests manually. I would like to see the % of tests that are messed up. I can't imagine with the number of tests taken each year that not a single one experiences any errors. Statistically speaking, you just can't have 200,000 tests given and each one be graded perfectly. From Pearson's perspective, they might say "Our error rate is less than 0.1%" or something low like that. That's great for Pearson, but what if you're one of those 200 people that got messed up because of the test? I'd be pissed! (angry, not drunk)

If I were you, I would contact Perason or GMAC and ask about their quality control measures. I wouldn't ask any specific questions, or even mention your own test results, but just inquire as to their quality control measures and what they do to test the accuracy of the grading system. That is something they should be able to tell you and don't settle for a simple explanation like "Our test results are accurate to 99.99% (whatever %)." I would ask for specific things the company does to measure quality of testing results. They have to have them in place and they should be able to disclose those. If they are hesitant about it, or say they can't disclose them. Start talking to other people. It may not change your test score, but it might help you accept the results, and thus, prepare effectively for the retake.

Did anyone called or emailed gmac/pearsonvue for revalution of verbal section score ? They might be able to cross check.

BTW I am not a native English speaker, but have done my entire education in English medium... On my GRE I scored on the 90th percentile in Verbal 6 years back.. On my practice tests for my GMAT I scored b/w 44-47 ( 27 twice, and 44 once.)

Now This is what surprises me... On the actual test I get 27. On all my practice tests the worst I got was a 35 on Kaplan.

On the score rechecking issue: Please let me know if thats possible with Pearson.

I suggest all of us who have scored low on Verbal, write a joint letter to GMAC, or Pearson asking them to look into this issue. Individual letters may not carry the same weight..

Again , it is highly possible that all of us have actually screwed up our test, but that all of us suddenly land on a 27, is really weird. Again this was my second attempt and in my first attempt I scored 35. That time I did not complete my Verbal and guessed Q 31 onwards. This time I had paced my self well, and really worked on my SC. Practiced LSAT questions and beefed up my CR. And I scored 27. Thats shocking...

Again on other forums, where this students have received shocking scores the GMAT Verbal score has been 27/28. This seems abnormal.

Unfortunately there really isn't anything they can manually grade, unless they had a video camera shooting which keys you entered for which question as you took the test.

The data is in the computer, they can't un-computerize it.

Also I presume there are about 20,000 GMATs given per day? Good luck on demanding manual QA on that.

There are way less than 20,000 given per day. If that were the daily total, that would equate to just under 6,240,000 exams per year. If there are 500,000 exams taken annually, that equates to about 1600 per day (6 days a week for 52 weeks)

The issue that occurred in 2001 was that certain questions in the middle of the quant section were recorded as unanswered. That's impossible for a test taker to do. It will not let you go to the next question unless you answer the current question. That is easy to find even though answers are in the computer.

Even if GMAC/Perason doesn't do manual QA, it would still be nice to know what methods are used to ensure the quality of the exam data generated.

Roughly 250,000 GMATs are taken annually.

I won't dismiss the possibility of some kind of scoring error, but you might be surprised by the QA built into the question-design process. I may be combining information that I've read about various tests, but the process works something like this: Questions are written, then edited, then attempted by a panel of 'expert test-takers', then if accepted, inserted as experimental questions into real tests, then the results are analyzed for differential item functioning (e.g. do Americans answer the question more successfully than non-Americans?), analyzed for test functioning (are the statistical properties of the question suitable for the test- is it too easy or too difficult, does it discriminate sufficiently between high and low level test-takers, etc). If the responses to a question are bizarre (e.g. if good test-takers answer the question incorrectly more often than bad ones), the question will be sent back to the start of the process (and this catches any 'wrong answers' recorded in the database). There's a reason it costs GMAC $2,400 to develop a single question.

I'd add that checking a GMAT score manually would just involve running the responses given back through the scoring algorithm; there's no way a person could do the calculations by hand.

Hi, IanStewart,

It seems that you know this industry very well. May I ask your opinion on the following issues?

a) Do you think it is probable that GMAC has recently changed the scoring algorithm for verbal?
b) In your opinion, is the scoring principle for verbal different from that for quant – and it is because of this the results of practicing tests in verbal are less indicative of the real score? Perhaps, the factor of chance is more important in verbal?

Thanks,
Greenoak.

As I read the forums and from my own experience I started to agree with you. I think the scoring algortihm must be changed or something. In my last exam I thought I nailed verbal but I got 38. If this is the case isnt it unfair. It puts people who took GMAT last year in an adventagous point.

Is anyone planning to approach GMAC to see if they can revaluate the score..